Oh....they said for the 9 XM-L model can use 3 x D sized 32650.
From KD's smaller 32600 5000mAh, mitro did 4500mAh @ 5A (I measured 5200mAh @ 1A). The KK 26650 did 10A nicely. So I think battery requirement would be taxed but it's not impossible. However it is non-protected, definitely not for the novices.
The light is the 2D80 model, 80mm++ head and 3D long. So something like TK70 just smaller by 20mm. 1kg mass, 200g more than TK70.
9-XML must use 3x 32650 tube. Not sure if it can do 3S3P 18650.
For the 7-XML and 5-XML models it can use 2D tube, or 2S3P 18650.
It's 80mm reflector with somewhat spaced XM-Ls. Hence it is still something like a floodlight, or think of it as slightly worse than 9 XM-Ls with individual reflectors squeezed inside. From the beamshots, it still throws less than the SR92, and the SR92 has 45k cd. It's damn bright (lumens) though.
PS. As always, never take the output specs at face value, someone tested the CSM-360 and was disappointed.
It's taxed real bad, but not that "mission impossible".
The problem with such a design is that it is not mounted on 14/16mm MCPCB like the DRY, TR-J12, usual triple XM-L suspects. Take for eg the DRY, ignore the output but basically that is a 40W or 40W plus light depending on the amps that the cells can push. It's mounted on a light weight host but the LEDs are on 14mm.
Even with direct emitter bonding, it's too close. Overall efficiency would be like the DRY once past a certain point (eg it's 3A for the triple XM-Ls and it slides down significantly). Even if the driver can take it, say instead of 8000 lumens theorectical, it probably might be 6500 lumens. With that mounting the efficiency takes a heat as it's really different from individual 14mm MCPCB or slightly cut 16mm MCPCBs.
6000-6500 emitter lumens from 9 XM-Ls sounds doable in such a package - but then again it's too expensive. 2 x TR-J12 gives nearly 5000 lumens OTF and costs about usd160 shipped.